Abstracts

Rationale: Intracranial electroencephalography (iEEG) is an indispensable tool for diagnosis and surgical planning in patients with Epilepsy. iEEG is measured using subdural grids or depth electrodes that can have a multitude of sizes and geometries. Therefore, it is critical to understand the precise relationship between electrode size and EEG measurement to better interpret and compare recordings from electrodes of different sizes. To address this, we devised a novel method to change the electrode surface area, and thereby size, after implantation in the human brain. Using data obtained via this method, we investigated the effect of electrode size on various signal features, including the morphology of high frequency oscillations (HFOs) and interictal spikes.

Methods: Three human subjects were each implanted with a high-density 8x8 subdural grid of intracranial EEG electrodes. After implantation, the effective electrode size was changed by electrically shorting adjacent electrodes in groups of two or four. This shorting effectively averaged the neural activity under the electrodes, thereby mimicking a larger surface. Twenty-minute iEEG recordings were thus obtained for three different electrode sizes from a single grid in a static brain location in each subject. The following properties were calculated for all three sizes: (1) The root-mean-squared (RMS) amplitude of the iEEG, (2) Correlation values for every pair of channels at a fixed inter-electrode distance, (3) HFO rate and morphological features, based on detection using an automated algorithm (Staba et al. 2002), and (4) Signal-to-noise ratio (SNR) of interictal spikes, based on manual marking of the iEEG data in the small electrode configuration. Statistical tests were then used to compare the features for different electrode sizes.

Results: The correlation between channels increased significantly with electrode size (p < 0.05) in the 30-60Hz and 60-120Hz frequency bands, while the amplitude of iEEG signals decreased with electrode size. The HFO rate per area of tissue decreased as electrode size increased, for both ripples (p < 10^-10) and fast ripples (p < 10^-6). The smallest electrode size recorded more fast ripples than ripples (p < 0.0001), contrary to what is typically reported for standard-size electrodes. HFO amplitude also decreased with electrode size, but duration and peak frequency were not affected. Across all visually marked spikes, 61% had the highest SNR when measured with the smallest electrodes, compared to 29% and 10% for the medium and large electrodes, respectively.